Hey guys,

I finally got an Osiris Tuner for my 03 G35. My G has a custom TT, a UTEC, DW 600cc injectors and a CJM stage 0 fuel return. Today I took a crack at some low load tuning to try to get the injectors scaled properly. I set the 0% column on the UTEC to all zeros, so the UTEC should not be interfering with anything when I'm not in boost. I was never happy with the way the UTEC handled the bigger injectors when not in boost. I always had little driveability and idle issues that I couldn't tune out. That's where the Osiris comes in.

I basically followed the guidelines in the Osiris user manual v2 on how to tune for the larger injectors. I started out by setting up the injector latency parameters for Deatschwerks 600's, which I found in a thread on here listed as Denso injectors (I sent an email to DW to see if they could send me their data, just to verify). Then I started lowering the K fuel multiplier to get my A/F compensations close to 100 at idle.

After that I did some low load, low rpm pulls with 12:1 as my target A/F in all zones. Doing this resulted in my AEM wideband reading full rich (10:1) for the majority of each pull, so I began lowering the K multiplier further until I could do a pull and the A/F would stay around 12:1 more or less the whole time. One thing that I dislike so far about the RTT Cipher software is that there are a lot of values in hexidecimal format, which makes it difficult to correlate them to actual numbers (Unlike the ROM editor, which shows most of the same numbers in "regular" decimal format). I resorted to using the calculator in Windows accessories to convert the numbers when I wanted to get an idea of how much I had changed them by. The other thing that didn't make sense right away is that the K multiplier is a huge number (on the stock ROM it was like 25,500!) so you have to change it by huge amounts to see any difference. At first I was timid about changing the value too drastically, but now I realize that one click on the up or down arrow only changes it by 1 point, and when the starting value is 25,000 that one point is not very significant. Eventually I just held the up or down arrow for about 30 seconds at a time when I wanted to make a change to the K. I ended up at about 16,000 for K after I made a few pulls.

However now that I have the K decently dialed in during low load, low rpm pulls, my idle A/F compensations are way high - like 125 or something. I am wondering why they are so far off? When I dial in the K at idle its like 18,300 - but when I dial it in while cruising its more like 16,000 so then it idles lean and the computer has to add a bunch of fuel to compensate. I can get the A/F compensations back to 100 by increasing the values in the fuel compensation table to like 118 at idle, but then they are way higher than the rest of the fuel compensation table so it seems weird to me.

My idle is not very stable either. It bounces back and forth from lean to rich (from about 16.5:1 to 13.5:1) and the idle speed surges between about 650 and 750. During this time the A/F compensations remain pretty stable in the 98-102 range (with 118% in that zone on the fuel compensation table). I also lowered minimum effective injector pulse from 2.4X ms to 1.50 ms (which is strange, because even when it was at 2.4 Cipher was showing an injector pulse width of ~1.8 at idle??). I am wondering if maybe my latency values are wrong?

More to come...

 

good info on here.... (subscribing)

keep us updated of your findings.

 

Well I just got a response from Mike at Deatschwerks after I emailed them regarding the correct injector latency for the Deatschwerks 600cc injectors. According to Mike the correct values are:

6v - 3.57
7v - 2.87
8v - 2.42
9v - 2.10
10v - 1.88
11v - 1.71
12v - 1.59
13v - 1.47
14v - 1.35
15v - 1.27
16v - 1.19
17v - 1.12

I assume these numbers are in milliseconds. These are waaay higher than what I was using before (.64ms at 14v and 1.150ms at 10v) which I got from this thread under Denso:

https://my350z.com/forum/forced-indu...injectors.html

This could explain why my idle A/F is lean while my cruising is not. The higher latency numbers would mean that I was not compensating enough for the latency, which would be especially noticeable at low pulse widths (idle).

I will test tonight with the new latency values and see what happens.

 

I got some more correspondence from Mike Deatsch after I remarked at how much larger the latency values he gave me were than what I had been previously using (and the stock latency values for that matter). Here's what he said:

What are you using for tuning? I would either start with the numbers I gave you or just the stock latency and adjust from here. You will probably have to play with the latency a little bit to get it right.

The value you listed, .68ms @14v is not a valid high impedance latency value. That is a latency value more in line with a low impedance injector.

The actual latency time of a injector and the numerical value uesed in tuning are not always one in the same...

That is not uncommon for software to show latency values such as you listed. That is not the actual latency time of the injector, rather just a representative value. There is probably some equation in the software that gets to that value. In this case, our latency values will not be of much help. You will just need to find ones that work best. I would start with stock latency and adjust from there as needed.

.48ms is a unrealistic actual latency value for a high impedance injector. Most high impedance injectors are between 1ms-1.5ms at 14v. Even low impedance injectors rarely have that low of latency time.

-Mike

Before I received the above emails I tried the latency values he gave me. The car ran super rich at first, until I lowered the K to about 8000 and reduced my minimum injector pulse to about 1.0 ms! After that it idled pretty well with A/F compensations at about 100. However this seems very strange to me that I would have to change the K multiplier so drastically. I didn't get a chance to do any low load tuning so I don't know how well the ECU would be hitting it's targets in all other zones...I suspect it would be lean.

Anybody got any good advice on how to properly tune injector latency? It's proving to be very difficult. I am about to go out and find a used two channel oscilloscope and measure it directly!

I also tried .85ms for a latency value at 14v, just to see what would happen. It seemed to be an improvement - when I tuned the K values for light load pulls and then used the fuel compensation to adjust the idle back to stoich I only had to go to about 103% at idle instead of 118% like before.

 

i wish i could help... i'm just in my learning phase (which will take years i suppose) on how to tune.

IMO, it might be difficult just like everyone said, Osiris isn't an easy EMS to tune with..

good luck on ur findings.. and i'm here following up

 

Well at least somebody cares! Thanks maXmood.

Honestly I figured that with all of the talk about Osiris on this form that more people would be interested in this discussion. I guess everybody wants to just plug it in and have it work. I wish the self-tuner base was bigger in the 350z community.

 

There's some good reading about Nissan ECU programming here:

http://ecu.ztechz.net/

Granted it talks about the 300zx ECU specifically but a lot of it is similar to the 350z ecu and it helped me to understand certain things like why some of the values in Cipher RTT are shown in hex!

There's also a nice little section in there about setting the new K value for larger injectors. By using the logic outlined by the author I get a new K of 16,300, which was very close to what I was getting during my initial tuning sessions.

 

^totally agree with you on that... we got lots of ppl talking about tuning and stuff, but can't see much of an interest in sharing those findings/ideas or even research... being an enthusiast, i think that what gives me the urge to learn even if it wasn't the best out there, because you can always build up on that.

thanks for the link.. i'm going through it this evening when i get the time..

on topic, i'll do a search on injector latency and let you know my findings (if any).

GL and keep us posted.

 

Hey MX594, sounds like you are on the right track and have a good foundation of knowledge. The K fuel value is very important...but dont drive your self trying to get it perfect cause in the end you will be using the MAF table to trim in your values for throttle tip in and WOT. Make sure you first set all your fuel comp tables to 100 that way you dont have them messin with ya to. They should be the last thing you use to trim in your fuel values.

Also if i could suggest one thing, i would go straight to a PMAS MAF cause your gonna max out that MAF volatge at about 350-375 HP. So when u are spending alot of time trimming in your MAF values...your gonna have to change all of them when you go to the PMAS MAF anyway. Hope this helps..

 

Thanks Vince!

So you are suggesting that I get a decent K value, then primarily tune the MAF curve and only use the fuel compensation table to make final fine tune adjustments? Fuel compensation is much easier to tune since it can be done in Cipher RTT.

I was under the impression that the MAF curve provided by UPREV for the stock MAF was very accurate since my MAF is still in the stock housing, so the curve should already be well defined for x voltage = y airflow. That is unless the piping ahead of my MAF (and the fact that it is blow though) changed the cross sectional flow profile enough to make the readings innacurate? This is where a draw through MAF setup would shine, since the flow characteristics would remain largely unchanged from the stock airbox setup (see JWT turbo kits).

I am sticking with the stock MAF for now and just trying to get a feel for the UPREV software. The UTEC takes care of the boosted operation for now. Until UPREV adds third party wideband datalogging support to Cipher I will probably keep the UTEC (I have the UTI software for the UTEC, which lets me log the UTEC data and combine it with the data stream from my AEM wideband). The UTEC sucked for driveability in vacuum though because of it's lack of ability to properly scale the injectors, which is where the Osiris comes in. Someday I would like to use Osiris to replace the UTEC completely but not until I understand it better.

After reading through that link I posted above, I tried the method that the author outlined for setting a K value. Basically I just took the "stock" K value and multiplied it by the percent difference in the stock vs. new injector size. So basically it was something like 27,300*(380/600) = 17,300.

Then I set out to determine approximately what the latency value should be, based off the data given to my by Deatschwerks and the base ROM provided by Uprev. In the Uprev software the base ROM listed latency at 14v = .48ms and at 10v = .84ms. Its a linear function so for discussion purposes I will jsut use 14v and 10v as the two data points. Now according to Deatschwerks the stock injector latency value at 14v = 1.27ms and at 10v = 1.77ms. I do not know why there is such a large discrepancy between what Deatschwerks has measured and what is listed in the base ROM provided by Uprev. Also according to Deatschwerks, the latency for their 600cc injectors at 14v = 1.35 and at 10v = 1.88. So I decided to take the difference and add it to the stock values listed in the base ROM from Uprev. So I ended up with 14v = .56ms and 10v = .95 ms.

I went ahead and plugged in these new latency values along with the 17,300 K I calculated, flashed the ecu and went for a drive. To my surprise it seemed to be hitting the A/F targets (14.7 in the zones I was hitting) almost perfectly! Idle still had A/F comensations over 100 so I added fuel with the fuel compensation table. It runs the best it has yet, and the idle seems to be much more stable.

The saga continues. Anyone get a two channel oscilloscope with a current clamp they want to sell cheap? I would like to meaure injector latency directly if possible.

 

good stuff!

keep me posted of any development.. and let me know when you move to tuning other stuff as well.

where can i find the parameters 10v and 14v? can't seems to find em..

 

Your def on the right track. The stock scale is very accurate for stock...but when u switch to a blow through set up it does get skewed a lil bit. Also i forgot you were still running the UTEC under boost conditions. Alot of the MAF scaling needs to be done under boost, so you should be good on that.

Make sure u also change your Minimum effective PW from 2.46(stock) to like .2ms under whatever your injectors are running at idle. I know its very tempting to just make all ur adjustments under the fuel comp tables but fuel comp tables are the last thing the ECU looks at for fuel trim, so sometimes you will find the ECU hunting a lil for the trims. I try to keep all my fuel comp tables between 95-105 so as long as you are near that u will be good!

 

maXmood,

In the ROM editor under utilities you will find an injector scaling utilitiy. This is where you specify the latency function and it will show you sample values at different voltages (v). The function is linear (it makes a straight line) but the latency values do not necessarily fall in a straight line. Since 10 volts (cranking) and 14 volts (running) are the most important points, those are the two points that you should fit the function to. Just play with the slope and the offset until you get the values you want to see at 10 volts and 14 volts.

Vince,

I reset all of my fuel values to 100 in the compensation table and set out to start mapping my own compensation table. Before doing this the table had lots of values in the 80's and 90's. In fact the highest value in the compensation table on the "Stock" ROM file was like 93.7. So after setting everything to 100 the car ran rich everywhere, so I had to lower the K to about 15,000 to get my crusing and light throttle A/F compensations close to 100. To get the idle A/F compensations back close to 100 I had to add a ton of fuel to the idle area of the fuel compensation table...like 114% or something like that.

Things are pretty out of wack now though, since my compensation table is not doing any "compensating". I started logging runs and looking at the O2 sensor readings and A/F compensations in an attempt to begin building a new compensation table for my engine. I generally would try to do a 3rd gear run from about 1500 rpm to 5000 rpm at various different Base Fuel Schedule values. For example I would do a run and try to hold the Base Fuel Schedule at 4 the whole run, then do another at 5 BFS, etc. Then I would add fuel in the fuel compensation table to areas where I saw lean O2 sensor readings and the fuel compensations going upward.

One thing that I have noticed with the Osiris and also when I was just using the UTEC is that I have a very lean spot right at about 2300 rpm and low load. I even have my fuel compensation table at 110% in the 2000 and 2400 rpm zones and the thing still studders when passing through this RPM. I can watch the AEM WBO2 readout go full lean for just a moment when it happens. I am wondering if I am maybe getting some flow reversion through the MAF, or if my MAF needs to be cleaned perhaps???

 

I forgot to add that my minimum pulse width is set to 1.5ms. Last time I checked I was seeing about 1.8ms at idle.

I might also add that third party wideband support would be absolutely fantastic!!! I have an '03 and therefore do not have factory wideband sensors. I can datalog the A/F compensations and the stock O2 sensor readings and that's what I have been using for tuning (since I am tuning in vacuum only, the stock NB sensor should be pretty accurate around 14.7 A/F). It sucks though becuase in the logs it doesn't show A/F but rather just O2 sensor voltage from 0-1v. There might be a way to specify a conversion table in Cipher to convert the 0-1v readout to an A/F ratio but I haven't tried yet.

 

^^^ yea when you set all the values to 100 it will prob wack things out a bit. Thats when you need to go in and modify MAF values to get the desired air fuels. Log MAF voltage and fuel trims and adjust accordingly in the MAF tables.. Trust me it will make things very smooth

 

OK, I'm convinced.

Ill leave the tuning to the professionals. Vince, you're in the right business.

MK594: Nice work so far. Nice to know that someone in the area has some knowledge on this. We need to setup a G & Z meet in Michigan!

 

I'm getting nowhere with this. I'm about ready to give up and buy a Haltech.

 

don't give up on me!!

 

Whats your issue?

 

I just can't seem to get it running right. I've tried making adjustments in the fuel compensation table as well as the MAF curve and everything I do seems to make things worse.

For example, I have a lean spot right around 2250 rpm. It seems to be worse at lower loads. It was there with the UTEC as well, and no amount of MAF votlage offset seemed to make a difference. Now that I have the Osiris I tried adding fuel in this area of the fuel compensation table and I also tried raising the MAF curve in the voltage range where I was seeing the lean spot during my logs but nothing seems to help. When I log the phenomenon I see my A/F compensations take a straight dive to 75 right at 2250 rpm, then they go straight up to 100+ as the rpm's climb. In other words when I log the A/F compensation it makes a big "V" with the center of the V right around 2250 rpm. So essentially the computer is pulling a ton of fuel and making it's own lean spot. Why it is doing this I have no idea. I tried cleaning the MAF with some MAF cleaner spray but it didn't help. The only thing I can think of is that my A/F sensor(s) are shot, but I would expect to see a code if that were the case.

Also, I see A/F compensations around 110 at idle and then all the way down in the 80's while driving. I have to make huge additions in the fuel compensation table (like 120%) at idle to get the A/F compensations close to what they are while driving. I tried to adjust the MAF curve but that was a disaster.

It's not like I have never done this before. I have road tuned three cars with speed density based stand alone systems and they ran great. The Uprev software is cumbersome and difficult to use, not to mention unrefined. It's like it is still in the development stages for crying out loud. I can't integrate WBO2 readings into my logs either, so I have to rely on the crappy stock NBO2 sensors. And what's the deal with not being able to log the knock sensor readout?

I know that if I get a Haltech the software will work, I will be able to datalog using dual WBO2 sensors, and the interface will be user friendly. For the hours I have spent screwing with Osiris and UTEC, the Haltech is starting to look affordable.